Traditionally Public Transport (PT) demand estimation relies on manual survey-based or, where available, smartcard passenger data. However, even if these are precious data for researchers, providers and municipalities make it rarely available. The problem is poised to worsen accounting for the variety of formats and the low granularity in which such data is available. Only recently first steps towards the use of more advanced ICT-based data-driven approaches have started to emerge, which can provide new opportunities for generating more data and insights into transit demand patterns and behaviour. In this paper, we tackle one specific instance of transit demand estimation, that of subway stations. We design and test the TransitCrowd tool, which estimates the transit users entering and exiting stations using as proxy the subway crowdness provided by Google Popular Times crowdsensed information that is available at sheer scale in any city. Then subway crowdness is mapped to the precise volume and temporal dynamics of entrances and exits profiles at the level of each subway station. TransitCrowd's key component is the one-time calibration which creates temporal signatures of the stations. We assess TransitCrowd's estimation accuracy for two cities across a two-months period, i.e., New York and Washington.